Objective Satellite-Based Detection of Overshooting Tops Using Infrared Window Channel Brightness Temperature Gradients
Total Page:16
File Type:pdf, Size:1020Kb
VOLUME 49 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY FEBRUARY 2010 Objective Satellite-Based Detection of Overshooting Tops Using Infrared Window Channel Brightness Temperature Gradients KRISTOPHER BEDKA,JASON BRUNNER,RICHARD DWORAK,WAYNE FELTZ, JASON OTKIN, AND THOMAS GREENWALD Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, Madison, Wisconsin (Manuscript received 19 May 2009, in final form 31 August 2009) ABSTRACT Deep convective storms with overshooting tops (OTs) are capable of producing hazardous weather con- ditions such as aviation turbulence, frequent lightning, heavy rainfall, large hail, damaging wind, and tor- nadoes. This paper presents a new objective infrared-only satellite OT detection method called infrared window (IRW)-texture. This method uses a combination of 1) infrared window channel brightness temper- ature (BT) gradients, 2) an NWP tropopause temperature forecast, and 3) OT size and BT criteria defined through analysis of 450 thunderstorm events within 1-km Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR) imagery. Qualitative validation of the IRW-texture and the well-documented water vapor (WV) minus IRW BT difference (BTD) technique is performed using visible channel imagery, CloudSat Cloud Profiling Radar, and/or Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) cloud-top height for selected cases. Quantitative validation of these two techniques is obtained though comparison with OT detections from synthetic satellite imagery derived from a cloud-resolving NWP simulation. The results show that the IRW-texture method false-alarm rate ranges from 4.2% to 38.8%, depending upon the magnitude of the overshooting and algo- rithm quality control settings. The results also show that this method offers a significant improvement over the WV-IRW BTD technique. A 5-yr Geosynchronous Operational Environmental Satellite (GOES)-12 OT climatology shows that OTs occur frequently over the Gulf Stream and Great Plains during the nighttime hours, which underscores the importance of using a day/night infrared-only detection algorithm. GOES-12 OT detections are compared with objective Eddy Dissipation Rate Turbulence and National Lightning Detection Network observations to show the strong relationship among OTs, aviation turbulence, and cloud- to-ground lightning activity. 1. Introduction and background weather satellite observation inferences (Setvak et al. 2008b), which has important implications for the earth’s An overshooting convective cloud top is defined by radiative balance and climate. the American Meteorological Society’s Glossary of Thunderstorms with OTs frequently produce hazard- Meteorology (Glickman 2000) as ‘‘a domelike protrusion ous weather at the earth’s surface such as heavy rainfall, above a cumulonimbus anvil, representing the intrusion damaging winds, large hail, and tornadoes (Reynolds of an updraft through its equilibrium level.’’ Overshooting 1980; Negri and Adler 1981; Adler et al. 1985; Brunner tops (OTs) indicate the presence of a deep convective et al. 2007). Thunderstorms with OTs are also often storm with an updraft of sufficient strength to penetrate associated with strong horizontal and vertical wind shear through the tropopause and into the lower stratosphere. and lightning through charge separation and accumula- OTs have been identified as localized sources of lower- tion in the storm updraft region (Ziegler and MacGorman stratospheric water vapor (WV) through cloud-resolving 1994; Wiens et al. 2005), both of which represent serious NWP modeling (Wang 2003; Chemel et al. 2008) and turbulence and safety hazards for in-flight and ground aviation operations. OTs also generate gravity waves as they interact with and penetrate through the tropopause, Corresponding author address: Kristopher M. Bedka, Coopera- tive Institute for Meteorological Satellite Studies, University of which can produce significant turbulence at large dis- Wisconsin—Madison, 1225 West Dayton St., Madison, WI 53706. tances from the OT (Heymsfield et al. 1991; Lane et al. E-mail: [email protected] 2003; Bedka et al. 2007). DOI: 10.1175/2009JAMC2286.1 Ó 2010 American Meteorological Society 181 182 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 49 Convectively induced turbulence (CIT) represents a CloudSat Cloud Profiling Radar profiles, and Cloud- significant hazard for the aviation industry. From 1992 to Aerosol Lidar and Infrared Pathfinder Satellite Obser- 2001, the Federal Aviation Administration (FAA 2004) vation (CALIPSO) cloud-top height retrievals from the shows that 4326 weather accidents occurred in the United National Aeronautics and Space Administration (NASA) States, of which 509 were cited as being related to tur- A-Train satellite constellation can be used to better un- bulence. Nearly 23% of these CIT-related accidents re- derstand the relationship between satellite OT inferences sulted in fatal injuries to the occupants of the aircraft and the physical cloud height (Stephens et al. 2002; (FAA 2004). FAA (2004) also indicates that the majority Mitrescu et al. 2008). (67%) of turbulence encounters over the United States Several published studies describe objective OT de- for general aviation occur during the warm season (April– tection techniques using multispectral satellite imagery. September). Cornman and Carmichael (1993) state that Berendes et al. (2008) show that a combination of visible CIT is responsible for over 60% of turbulence-related and near-IR texture and reflectance, IR channel BTs, aircraft accidents. Because of the hazards associated with and multispectral IR channel BT differences (BTD) can OTs, objective detection of OTs is a product requirement be used in an unsupervised cloud classification tech- for the Geostationary Observing Environmental Satellite nique to objectively detect deep convection and OTs. (GOES)-R Advanced Baseline Imager (ABI) instrument While this technique performs well at and around the program (Schmit et al. 2005). The GOES-R is currently hours near solar noon, Berendes et al. show a signifi- scheduled for launch in 2015. cant diurnal signal in classifier output that can be in- Signatures in multispectral weather satellite imagery duced by enhanced texture in visible channel imagery indicate the presence of OTs. OTs exhibit a lumpy or atop deep convective clouds at low solar zenith angles. ‘‘cauliflower’’ textured appearance in visible channel im- A close inspection of OT detections during the early agery. OTs are also inferred through the presence of morning and evening hours with corresponding visible a small cluster of very cold brightness temperatures and IRW channel imagery suggests that many of these (BTs) in the ;11-mm infrared window (IRW) region. pixels would be considered false alarms. Near-IR re- OTs continue to cool at a rate of 7–9 K km21 as they flectance and ice particle effective radius techniques ascend into the lower stratosphere (Negri 1982; Adler (Lindsey and Grasso 2008; Rosenfeld et al. 2008) suffer et al. 1983), making them significantly colder than the from some of the same issues, making these techniques surrounding anvil cloud temperature. The surround- unreliable for and/or not applicable to objective day/night ing anvil cloud has been shown to have temperatures OT detection. at or near that of the tropopause level (Adler et al. As the GOES-R ABI satellite program and opera- 1985). tional forecasters require product availability during both Techniques and instrumentation such as numerical day and night, the use of infrared-only techniques for modeling, aircraft photography, multisatellite stereos- objective OT detection has also been investigated. The copy, active space-based radar, and aircraft-based lidar 6–7-mm water vapor absorption minus the ;11-mm IRW have been used to better understand the relationship channel BT difference technique for OT detection has between OT signatures in visible and IRW satellite im- been described extensively in the literature (Fritz and agery and the physical height of the cloud top. From sev- Laszlo 1993; Ackerman 1996; Schmetz et al. 1997; Setvak eral flights over OTs with an airborne lidar, Heymsfield et al. 2007; Martin et al. 2008). The premise behind the et al. (1991) showed that some OTs reach altitudes up to use of this technique for OT detection is that 1) the at- 2 km above the surrounding anvil cloud. From Tropical mospheric temperature profile warms with height in the Rainfall Measuring Mission (TRMM) precipitation radar lower stratosphere, 2) water vapor is forced into the data (Kummerow and Barnes 1998), Liu and Zipser lower stratosphere at levels above the physical cloud top (2005) found an overshooting magnitude of 0.67 km for by the overshooting storm updraft, 3) this water vapor global deep convective clouds. The height difference emits at the warmer stratospheric temperature whereas between lidar and TRMM-based results is likely due to emission in the IR window channel originates from the the fact that lidar derives cloud-top height via ice crystal colder physical cloud top, and 4) positive differences reflectance whereas TRMM requires reflectance from between the warmer WV and colder IRW BTs can precipitation particles that reside at lower levels within therefore identify where overshooting is occurring. The the cloud. Liu and Zipser also show that overshooting maximum WV-IRW BTD can be offset from the OT magnitude depends significantly on whether the level